New research that improves prospects for prevention or treatment of ovarian cancer is essential to reduce the burden of this disease. Although only one-eighth as common as breast cancer, ovarian cancer accounts for a disproportionately large number of deaths, due to its typical presentation in an advanced stage with little chance for cure. Epithelial ovarian cancer is now considered not as a single disease, but rather as a diverse group of tumors with subtypes that can best be classified based on molecular genetic features. We will apply this model to assess the association of tumor subgroups with known or suspected ovarian cancer risk and preventive factors and with disease outcome. As much as 75% of epithelial ovarian cancer is now regarded as high-grade serous (HGSC), and accounts for 90% of disease mortality. This provides strong incentive to employ novel methods to identify and assess biologically relevant subgroups of HGSC. Identifying subtypes with true etiologic differences has important implications for prevention and for improved, targeted therapy. In the proposed study, we will follow up on intriguing findings of The Cancer Genome Atlas Research Network and related research that has identified four robust subtypes of HGSC based on patterns of mRNA expression. In two population-based studies of 2240 invasive ovarian cancer cases and 2900 controls (with detailed information on reproductive, lifestyle and medical histories, and on germline genetic variation), we propose to: 1. a. Classify tumors as HGSC, low-grade serous (LGSC), endometrioid (EC), clear cell (CCC) or mucinous (MC), using protein (IHC) and mRNA (NanoString) based classification schemes; b. Sub-classify HGSC into four robust and reproducible subgroups according to mRNA expression patterns, and describe the prevalence of each subtype in our population-based samples; 2. Examine whether associations with known or putative epidemiologic and genetic risk and protective factors differ by protein and mRNA expression subtype; 3. Examine whether survival differs by protein and mRNA expression subtype An integral strength of our approach is the examination of novel, molecularly-defined and biologically meaningful subtypes of epithelial ovarian cancer. We will use the NanoString nCounter platform, a highly sensitive and accurate multiplex assay, to directly measure mRNA expression levels from representative sections of formalin-fixed paraffin-embedded tumor specimens. Notably, we will examine epidemiologic differences across four subgroups of HGSC, which has not previously been done. Our findings can importantly influence the development of more effective strategies for disease prevention and treatment.